Assuming that alignment by default happens with nontrivial probability, one way to produce human-aligned AGIs would be to simultaneously create sufficiently many (different) AGI. This leads to a multipolar scenario where there are a few AGI aligned with humans, and many unaligned AGI. (By unaligned I mean unaligned with humans,...
Suppose that next year, AI Safety is solved, the solution is approved by Eliezer Yudkowsky, etc. How do we actually get people to follow this solution? It seems to me a lot of people or companies will ignore any AI Safety solutions for the same reasons they are currently ignoring...
Given an aligned AGI, to what extent are people ok with letting the AGI modify us? Examples of such modifications include (feel free to add to the list):
What exact parts of being "human" do we want to preserve?
A "moonshot idea" I saw brought up is getting Yudkowsky's Harry Potter fanfiction translated into Chinese (please never ever do this).
This has already been done, and has pretty good reviews and some discussions.
I've looked through the EA/Rationalist/AI Safety forums in China
If these are public, could you post the links to them?
there is only one group doing technical alignment work in China
Do you know the name of the group, and what kinds of approaches they are taking toward technical alignment?
I'll use this comment to collect things I find.
Are there any alignment approaches that try to replicate how children end up loving their parents (or vice versa), except with AI and humans? Alternatively, approaches that look like getting an AI to do Buddhist lovingkindness?
For derendering latex in Emacs, see https://github.com/io12/org-fragtog.
For drawing images in line, you could try https://github.com/misohena/el-easydraw.
I like this idea and think it is worth exploring. It is not even just with training new models; AGI have to worry about misalignment with every self-modification and every interaction with the environment that changes itself.
Perhaps there are even ways to deter an AGI from self-improvement, by making misalignment more likely.
Some caveats are:
Assuming that alignment by default happens with nontrivial probability, one way to produce human-aligned AGIs would be to simultaneously create sufficiently many (different) AGI. This leads to a multipolar scenario where there are a few AGI aligned with humans, and many unaligned AGI. (By unaligned I mean unaligned with humans, these AGI may or may not be aligned with some other goal.)
Although it is true that in general having many AGIs is not good, the idea is if some of those AGIs are aligned, this may be a better outcome than having just 1 AGI that is probably not aligned. Or perhaps, once there are enough aligned AGI, since they are all... (read more)
Escape. Invest in space travel and escape the solar system before they arrive.
If your AI timelines are long, this may be a viable strategy for preserving the human species in the event of unaligned AGI.
In your AI timelines are short, a budget solution is to just send human brains into space and hope they will be found and revived by other powerful species (hopefully at least one of them is "benevolent").
Suppose that next year, AI Safety is solved, the solution is approved by Eliezer Yudkowsky, etc.
How do we actually get people to follow this solution?
It seems to me a lot of people or companies will ignore any AI Safety solutions for the same reasons they are currently ignoring AI Safety:
- They think AGI is still very far away, so AI Safety methods won't need to be applied to the development of current narrow AI systems
- The concepts of AI Safety are difficult to understand, leading to incorrect application of an AI Safety solution or failure to apply it at all
- They will fall behind their competitors or make less money if they adhere to the AI Safety solution
- They have a new cool idea they want to test out and just don't care about or believe in the concerns raised by AI Safety
Thoughts?
I would be interested in the list of your favourite links!